A demographic comparison of two southern elephant. seal populations. CLIVE R. MCMAHON*, HARRY R. BURTON* and MARTHÁN N. BESTER

Transcription

1 Ecology 003 7, A demographic comparison of two southern elephant Blackwell Science, Ltd seal populations CLIVE R. MCMAHON*, HARRY R. BURTON* and MARTHÁN N. BESTER *Australian Antarctic Division, Channel Highway Kingston, 7050, Tasmania, Australia; and Mammal Research Institute, Department of Zoology and Entomology, University of Pretoria, Pretoria, Gauteng, Republic of South Africa Summary 1. To estimate concurrent age-specific survival for southern elephant seals at Macquarie and Marion islands, seals were marked from 1993 to 1997 in the first 3 weeks of life and resighted (recaptured) on return to their natal islands ( ). These recaptures formed the basis for the survival analysis in the mark recapture program MARK. Weaning masses were collected at each location.. Recapture probabilities were ( χ 6 = , P < ) higher at Marion Island than at Macquarie Island. There are two possible reasons: (1) the population at Marion Island is smaller and less dense than at Macquarie Island and () seals hauled out along a smaller section of the coast at Marion Island than at Macquarie Island, which: (1) facilitates the detection and individual identification of seals and () increases access to hauled out seals. 3. Age-specific survival estimates (corrected for preweaning mortality and tag loss) differed ( χ 5 = 64, P < 0 05) at the two islands and were consistently higher at Macquarie Island. The survival estimates for male and female seals were different at Macquarie Island ( χ 6 = , P < ) and Marion Island ( χ 6 = 0 373, P = 0 00). Female survival estimates were higher than male survival estimates. The combined survival estimates for juvenile seals (1 3 years) differed between islands but survival of older seals (4 6 years) did not. The inclusion of gender in the survival models did not improve model performance and hence male and female estimates were considered jointly. 4. The mean wean masses of male and female seals combined from 1993 to 1998 were not different between islands (T 6837 = 1 169, P = 0 4). At Macquarie Island the mean annual wean mass was kg (SD = 7, n = 6504) while at Marion Island it was 10 6 kg (SD = 4 7, n = 335). 5. The mean age at first breeding was different (P < 0 001) at the two island populations. At Macquarie Island the mean age of first breeding was 4 68 years, and at Marion Island it was 3 95 years. More ( χ 1 = 67 39, P < ) 3-year-old females breed at Marion Island (8 7%) than at Macquarie Island (1 %) and the proportion of seals that had bred at least once by age 7 was greater at Marion Island than at Macquarie Island. 6. We conclude that the observed decreases in elephant seal numbers between the 1950s and 1990s in the Pacific and Indian Ocean sectors were driven principally by resource limitation in the Southern Ocean. A conglomerate of factors including local predation by killer whales and intraspecific resource competition is postulated as a cause for the inter-island (regional) differences in population trends. It appears that more resources are available per capita to the Marion Island population than are available to the Macquarie Island population. Key-words: environmental change, mark recapture, sub-antarctic, survival, wean mass. Ecology (003) 7, 003 British Ecological Society Correspondence: Clive McMahon and Harry Burton, Australian Antarctic Division, Channel Highway Kingston, 7050, Tasmania, Australia. Clive.McMahon@aad.gov.au, Harry.Burton@aad.gov.au

2 6 C. R. McMahon, H. R. Burton & M. N. Bester 003 British Ecology, 7, Introduction Southern elephant seals (Mirounga leonina, Linnaeus 1758) have a circumpolar distribution and four distinct population stocks are recognized: the Peninsula Valdés stock in Argentina, the South Georgia stock in the South Atlantic Ocean, the Kerguélen stock in the south Indian Ocean and the Macquarie stock in the southern Pacific Ocean (Slade et al. 1998; Hoelzel, Campagna & Arnbom 001). The principal breeding colonies are located at: Peninsula Valdés, South Georgia Island, Heard Island, Îles Kerguélen and Macquarie Island. Overall, the southern Pacific and southern Indian Ocean southern elephant seal populations have decreased over the last few decades (Burton 1986; Condy 1978; Hindell & Burton 1987; Hindell 1991; Guinet, Jouventin & Weimerskirch 199; Pistorius, Bester & Kirkman 1999a). Recently the decrease at Îles Kerguélen was believed to have reversed (Guinet et al. 1999), and to have stabilized at Heard Island (Slip & Burton 1999) and Marion Island (Pistorius et al. 001), but continues at Macquarie Island (Hindell, Slip & Burton 1994). The southern Atlantic Ocean populations at South Georgia, the Falkland Islands and Peninsula Valdés have remained stable or increased (Campagna, Lewis & Baldi 1993; Boyd, Walker & Poncet 1996; Galimberti & Boitani 1999), with only the very small population at Gough Island showing a longterm decrease (Bester et al. 001). The cause(s) for the decrease in the southern Pacific Ocean population remain(s) unknown, although it may be similar to that at Marion Island where food limitation has been implicated (Pistorius, Bester & Kirkman 1999b). Long-term mark recapture studies were established at Macquarie Island (McMahon, Burton & Bester 1999) and at Marion Island (Bester 1988; Pistorius et al. 1999b) to address the proximate causes of the observed decreases. In both studies, elephant seals were marked and followed by subsequent recaptures, to determine agespecific survival. Survival (or mortality) is probably the most useful demographic parameter that enables ecologists to interpret and understand animal population dynamics for fundamental and applied purposes (Lebreton et al. 199; Lebreton, Pradel & Clobert 1993). Survival is important because changes in survival are often associated with changes in population structure and size (Smith & Fowler 1987; Lebreton et al. 1993). Age-specific survival estimates are especially important (Caughley 1977). To determine age-specific survival animals need to be marked permanently and individually, or marked in a fashion that allows for compensation of lost marks, and then followed throughout their lives (Caughley 1977). This requires long-term monitoring of the population and a method of marking animals that essentially is permanent, legible and has no mortality effects. Permanent emigrations from a birth site where animals are marked are important behavioural traits when estimating survival rates because animals that do not return to their natal sites are not available for recapture (Caughley 1977). Emigration from a capture mark recapture study site therefore compromises survival through wrongly assigning live animals to a dead class through non-capture (Caughley 1977; Lebreton et al. 1993). There is little permanent migration of elephant seals to and from Macquarie Island and Marion Island (Nicholls 1970; Condy 1978; Bester 1988; Slade et al. 1998; Hindell & McMahon 000). Slade et al. (1998) estimate exchange to be approximately females per generation between most populations. We assumed therefore that loss of seals through migration is negligible. By studying two populations, one that continues to decrease (Macquarie) and one that has possibly stabilized (Marion), we aim to identify characteristics of both populations as a means of elucidating the causal parameters that may be driving/ have driven the population decreases. To achieve this we: 1. Assess the concurrent age-, sex- and cohort specific survival of the two populations.. Determine the mean wean masses of seals at each location as a proxy measure of maternal foraging success and prey availability. 3. Calculate the mean age at first breeding for each site. 4. Propose a hypothesis that describes the causal factors driving elephant seal decreases. Methods MARKING AND RESIGHTING AT MACQUARIE ISLAND At Macquarie Island 43 recently weaned and tagged southern elephant seal pups were captured and branded from 1993 to 1997 on the isthmus (54 30 S, E). Fifty-millimetre cast-iron cattle brands were used to hot-brand seals on both rear flanks (Carrick & Ingham 1960, 196; Chittleborough & Ealey 1951). A fourcharacter alphanumeric brand consisting of a letter prefix followed by a three-digit number uniquely identified the cohort and the individual, respectively. All 43 branded seals were tagged at birth in the interdigital webbing of their hind flippers with two uniquely numbered plastic tags (McMahon et al. 1997). Daily searches of the isthmus beaches and tussock areas and monthly searches of the entire island beaches and tussock areas were made to resight (recapture) marked seals from 1994 to 001. At the time of resighting, location on the island, sex, brand number, tag numbers and the number of tags present were recorded. Both brands and tags were read. Flipper tags were used to validate brand identifications and vice versa. MARKING AND RESIGHTING AT MARION ISLAND At Marion Island (46 54 S, E), 056 recently weaned seals were individually marked with two

3 63 Elephant seal life history and population change 003 British Ecology, 7, uniquely numbered and colour-coded plastic tags in their hind flippers from 1993 to 1997 (Pistorius et al. 000). Beaches occupied by elephant seals were searched every 10 days except for the breeding seasons, when they were searched every 7 days (Pistorius et al. 000). The seal marking and resighting techniques used at Marion Island are reported in full elsewhere (Pistorius et al. 1999b, 000). They differ from those at Macquarie Island only in that tags are applied at weaning and not at birth. Cumulative age-specific tag retention rates, estimated from double-tagged individuals (Pistorius et al. 000), were used to adjust the survival estimates to compensate for tag loss. WEAN MASS MEASUREMENTS A total of 43 pups were flipper tagged at birth and weighed (±1 kg) on the day of weaning at Macquarie Island after McMahon et al. (1997). Some 335 pups at Marion Island were marked on the day of weaning and were weighed (±1 kg) shortly (0 days) thereafter, and their reconstituted weaning masses calculated following Wilkinson & Bester (1990). These wean masses were compared to those presented by Burton et al. (1997). The wean masses were included as individual covariates in a mark recapture model to assess the effect of maternal investment on age-specific survival, because wean mass acts as a proxy for maternal investment (Arnbom, Fedak & Boyd 1997; Fedak, Arnbom & Boyd 1996) and probably prey availability as well (Burton et al. 1997). Individual covariates of survival such as wean mass were included in the survival model by expressing the natural logarithm of the probability of survival, i.e. the logit of survival, as a logistic function of the covariates: x x Logit ( φ) = y-int ercept + β( x) β ( x ) SDx x x SDx eqn 1 where Logit (φ) is the survival estimate of a seal with the covariate x, β is the logit function parameter calculated in MARK for covariate x and SD is the standard deviation of the covariate x. This model (function) is embedded in the log-likelihood function for survival as in a logistic regression. This model assumes that there is an optimal value for the variable x and that there are some selective penalties associated with the extreme values of x. MARK RECAPTURE ANALYSIS Capture-history matrices were constructed from the resight histories of individual seals. Multiple resights within a year were treated as a single sighting. These capture matrices were used as input files for the capture mark recapture (CMR) program MARK (White & Burnham 1999) to estimate survival and capture probabilities after weaning. MARK provides survival (φ) and recapture (ρ) estimates under the Cormack Jolly Seber (CJS) model (Cormack 1964; Jolly 1965; Seber 1965) and under several models that appear as special cases of the CJS model (Lebreton et al. 199). Parametric goodness-of-fit (GOF) tests within MARK were used to test whether the CJS model assumptions were met (Burnham et al. 1987; Lebreton et al. 199). This bootstrap procedure simulates encounter histories that exactly meet the CJS model assumptions. These simulated data are compared to the field data for compliance with the CJS model assumptions (White & Burnham 1999). To test the main hypothesis (e.g. effect of sex, age and cohort on survival) the χ likelihood ratio test (LRT) statistics within program MARK were used (Lebreton et al. 199; White & Burnham 1999). MEAN AGE OF PRIMIPARITY AND NET REPRODUCTIVE RATIO The mean ages at which females gave birth for the first time (primiparity), at Macquarie Island and at Marion Island, were calculated using the model described by demaster 1981). This model calculates the probability of a female giving birth at a particular age from the number of females seen with young and the total number of females seen in that age class. Because breeding and moulting occur at different times and not all living females haul out during the breeding season, the total number of seals known to be alive at that age was used to represent the total number of females in that age class. All females hauling out during the breeding season were assumed to give birth (Pistorius et al. 001). An often used population fitness measure is the per generation ratio of increase (multiplication), R 0, that is defined as follows: R0 = lx m eqn where l x is the proportion of individuals that survive from birth to age x, and m x is the average number of female offspring produced by a female at age x. This ratio of multiplication was calculated and extended for each island population up to and including females aged 10 years by assuming that survival and fecundity were constant for adult females (Hindell 1991; Pistorius et al. 1999a; Pistorius et al. 001). Results GOODNESS OF FIT x The parametric GOF bootstrap (MARK) results for both the Macquarie Island and Marion Island data sets show significant (P < ) departures from the CJS model assumptions. The overall data sets were analysed further using program RELEASE to explicate the causes for the lack of fit (Burnham et al. 1987). The sum of the overall, male and female Macquarie Island

4 64 C. R. McMahon, H. R. Burton & M. N. Bester Fig. 1. The recapture probability estimates (±95% confidence limits) of southern elephant seals for Macquarie (solid line) and Marion (dashed line) Islands. The recapture rates were calculated in program MARK. Table 1. Selection of the most parsimonious models for the Maqcquarie and Marion Island data sets under the full CJS model for estimating survival in elephant seals. Model AICc AICc AICc weight NP Deviance φ (age, island group, log (wean mass) ages 1 at MQ, linear (wean mass) age 1 at MR) ρ (island group, time) φ (age, island group, log (wean mass), ages 1 at MQ and age at (MR) ρ (island group, time) φ (age, island group, log (wean mass) at MQ and linear (wean mass) at (MR) ρ (island group, time) φ (age, island group, log (wean mass)) ρ (island group, time) < φ (age, log (wean mass) ρ (island group, time) < φ (age, island group, log (wean mass), ages 1 MQ, ages (MR) ρ (island group, time) φ (age, island group, linear (wean mass), ages 1 MQ, ages 1(MR) ρ (island group, time) φ (age, linear (wean mass)) ρ (island group, time) φ (age, island group) ρ (island group, time) φ (age) ρ (island group, time) φ (age) ρ (time) φ (time, island group) ρ (island group, time) British Ecology, 7, seals, χ -value for Test was (d.f. = 0, P < ) indicating significant variation in recapture rates of seals. The overall χ -value for Test 3 was (d.f. = 16, P < ) that is indicative of significant variation in survival. Similar results demonstrating heterogeneous recapture and survival rates were obtained for the Marion Island population; the Test χ -value was (d.f. = 10, P < ) and the Test 3 χ -value was (d.f. = 18, P < ). Because heterogeneity in capture and survival probabilities (departure from the CJS model assumptions) have been shown not to affect survival rate estimates substantially (Carothers 1979; Nichols et al. 198; Pollock & Raveling 198; Barker 199), we present our survival estimates as those calculated from program MARK. RECAPTURES Recapture probability varied significantly between years at both Macquarie Island ( χ 5 = 01 3, P < ) and at Marion Island ( = 185 8, P < ) χ 5 (Fig. 1). The recapture probabilities of marked seals at Marion Island were significantly ( χ 6 = 376 5, P < ) higher than those recorded at Macquarie Island (Fig. 1). The most parsimonious models for the Macquarie Island and Marion Island data sets were those that incorporated age- and cohort-based survival components and time- and age-based recapture components (Table 1). The inclusion of a gender element in the recapture component did not improve the model significantly ( χ 15 = 4 1, P = 0 063). However, gender significantly ( χ 0 = 48 55, P = ) affected the survival estimates at Macquarie Island and at Marion Island ( χ 0 = 3 9, P = ). Female seals at both locations had consistently higher survival rates than male seals. AGE-SPECIFIC SURVIVAL ESTIMATES Significant ( χ 5 = 6, P < ) differences were evident in the age-specific survival estimates of seals at

5 65 Elephant seal life history and population change Fig.. The age specific survival rate estimates (±95% confidence limits) for southern elephant seals at Macquarie (solid line) and Marion (dashed line) Islands. The survival estimates are for male and female seals combined. Fig. 3. Concurrent age specific survival estimates (±95% confidence limits) for male and female southern elephant seals at Macquarie Island and at Marion Island for ages British Ecology, 7, Macquarie and Marion Islands (Fig. ). The most notable differences between the populations occurred in the first, second and fifth years of life. At Macquarie Island survival estimates for all years, except the third year, were higher than at Marion Island (Fig. ). Overall age specific survival for male seals at Macquarie Island differed significantly from that for males at Marion Island ( χ 6 = 4 46, P = ), the most notable differences occurring during the first year of life. Similarly for females, overall age-specific survival differed between the islands ( χ 6 = 34 37, P < ) and the major differences occurred in the first and second years of life (Fig. 3). The mean first-year survival estimate for five successive cohorts at Marion Island was significantly (Z 5 = 3 87, P < 0 001) less than for concurrent cohorts at Macquarie Island (Fig. 4a,b,c). The variance in firstyear survival for the five Marion Island cohorts was , while it was for the five Macquarie Island cohorts. Age-specific survival estimates for male and female seals were significantly different at both Macquarie Island ( χ 6 = 34 66, P < ) and at Marion Island ( χ 6 = 0 37, P = 0 00). Female survival estimates were higher than male survival estimates. Inclusion of sex effects did not contribute significantly to model performance when wean mass was included in the models for Macquarie Island seals ( χ 10 = 11 59, P = 0 31) or Marion Island seals ( χ 8 = 4 54, P = 0 81). Thus, male and female seals were pooled in the age-specific survival analysis in order to assess the impact of wean mass. WEAN MASS AND SURVIVAL The mean wean masses of male and female seals combined from 1993 to 1998 were not significantly different between islands (T 6837 = 1 169, P = 0 4). At Macquarie Island the mean wean mass was kg (SD = 7, n = 6504) while at Marion Island the mean wean mass was 10 6 kg (SD = 4 7, n = 335). Elephant seal wean masses during this study ( ) at Macquarie Island were similar (P = 0 94) to those collected earlier ( ) (Burton et al. 1997), kg (SD = 7, n = 6504) and kg (SD = 0 7, n = 463), respectively.

6 66 C. R. McMahon, H. R. Burton & M. N. Bester Fig. 4. The concurrent first-year survival estimates (±95% confidence limits) for five cohorts of Macquarie and Marion Island seals between 199 and 1998, for: (a) male and female seals combined (b) female seals and (c) male seals. 003 British Ecology, 7, However, the weaning masses at Marion Island were significantly different for these two periods (P = ) at 10 6 kg (SD = 4 7, n = 335) and kg (SD = 0 6, n = 411), respectively. The most parsimonious survival models at each island included wean mass as a covariate (Table 1). Wean mass influenced survival differently at the two island locations ( χ 16 = 38 66, P = 0 001) (Fig. 5). Wean mass influenced first- and second-year survival at Macquarie Island but only first-year survival at Marion Island (Table 1), and this model performed better than the model that included wean mass as a covariate for all age-specific survival estimates ( AIC 0 0 vs. 8 3). Moreover, at Macquarie Island first-year survival was a polynomial function of wean mass and was described most accurately by the function; y (survival) = (wean mass) (wean mass) that accounted for more than 99% of the observed variation in first year survival (r = 0 997). The relationship between wean mass and survival probability at Marion Island was

7 67 Elephant seal life history and population change best described by the exponential function y (survival) = e0 0071(wean mass) which accounted for 99% of the observed variation in first year survival (r = 0 991). MEAN AGE OF PRIMIPARITY AND NET REPRODUCTIVE RATIO The mean age at first breeding differed (P < 0 001) between island populations. At Macquarie Island the mean age of first breeding was 4 68 years ± 0 38, and at Marion Island it was 3 95 years ± A significantly ( χ 1 = 67 39, P < ) greater proportion of 3-yearold females breed at Marion Island (8 7%) than at Macquarie Island (1 %) and the proportion of seals that have already bred at least once by age 7 is greater at Marion Island than it is at Macquarie Island (Fig. 6). The net reproductive ratio up to age 10 was lower at Marion Island than at Macquarie Island (Table and Table 3), despite earlier initiation of breeding at Marion Island. While these differences were greatest when it was assumed that all the females in a specific age class produced offspring (Table 3), the difference persisted even when the actual proportion of breeding females (to age 7, and extended to age10) was used in the calculation of R 0 (Table 3). JUVENILE AND EARLY ADULT SURVIVAL Fig. 5. The relationship between elephant seal wean mass and survival in the first 3 years of life. The combined survival estimates for juvenile seals (1 3 years) differed between islands, but survival of older seals (4 6 years) did not (Fig. 7). Overall there were significant differences in survival between the islands ( χ = 7 344, P = 0 054) with survival of elephant seals (ages 1 7 years) being greater at Macquarie Island than at Marion Island. Survival of older animals (4 7 years) was greater than juvenile survival at both islands (Fig. 7). 003 British Ecology, 7, Fig. 6. The total proportions of female seals breeding at Macquarie and Marion islands and the proportion of seals that breed for the first time in the first 7 years of life.

8 68 C. R. McMahon, H. R. Burton & M. N. Bester Table. A life-history table for female elephant seals to age 10 at Marion Island and at Mcquarie Island. Included in the table are the survival probabilities p x from age x to x + 1-values, the proportion l x of individuals that have survived from birth to age x, and the mean number of female offspring m x produced by a female seal while in a particular age class. The net reproductive rate R 0 to age 10 is presented for each population and defined as the Σl x m x. The annual rates of change for the two populations are 4 8% per annum and 1 48% per annum at Marion Island (Bradshaw et al. 00) and at Macquarie Island (AAD records), respectively Marion Island elephant seals Macquarie Island elephant seals Age p x l x m x l x m x p x l x m x l x m x R 0 = Σ l x m x = R 0 = Σ l x m x = Table 3. The adjusted net reproductive rate R 0(adj) to age 10 for female southern elephant at Marion Island and at Macquarie Island taking into account the actual proportion of seals known to have bred in each age group. R 0(adj) is defined as Σ(l x m x ) b x where b x is the proportion of breeding females in each age group Marion Island Macquarie Island Age l x m x b x (l x m x )b x l x m x b x (l x m x )b x R 0(adj) = Σ(l x m x ) b x = R 0(adj) = Σ(l x m x )b x = British Ecology, 7, Fig. 7. Concurrent juvenile and adult survival estimates (±95% confidence limits) for juvenile (1 3 years) and adult (4 6 years) southern elephant seals at Macquarie Island and at Marion Island. Juvenile survival estimates were lower than adult survival estimates at both locations.

9 69 Elephant seal life history and population change 003 British Ecology, 7, Discussion The recapture rates of marked seals were higher at Marion Island than at Macquarie Island over the first 7 years of life. These differences are likely to be a function of the much smaller pup production (n = 430 pups) at Marion Island compared to Macquarie Island (n = pups) so that resightings in harems and moulting aggregations were easier to obtain at Marion Island. Although concentrated in the Isthmus area, seals were dispersed around the entire 96 km coastline at Macquarie Island while at Marion Island the seals only occurred along the coast in the main study area (5 km), and were generally absent from the west coast. GOODNESS OF FIT Non-compliance with the CJS assumptions, in particular the assumption that all animals have the same probability of survival, may have influenced survival rates but it is more likely to have impacted on the SE of these estimates (Carothers 1979). Although size, and therefore condition, at weaning influenced first-year survival rates at Macquarie Island (McMahon, Burton & Bester 000), heterogeneity in capture probability is more likely to be the reason for the departures from homogeneous survival probabilities (Pistorius et al. 00a). This is because juvenile seals are less philopatric than adults (Hofmeyr 001). However, the CJS model is biologically sensible. This is because survival varies annually (Pistorius et al. 1999b), larger weaned seals have greater probabilities of survival (McMahon et al. 000) and haulout behaviour and hence recapture probabilities vary with age (Hindell & Burton 1988; Hindell 1991) and season (see Pistorius et al. 00a). JUVENILE SURVIVAL First-year survival estimates for each of the five cohorts were more variable at Marion Island than at Macquarie Island and may indicate greater differences in female foraging success (prey distribution) and acquisition (prey availability) of resources (which influence maternal energy stores and transfer from mothers to their pups) for Marion Island females. This variability is related to maternal success because postnatal growth and survival are partly determined by the reserves a mother elephant seal has stored prior to breeding to provision her pup (Fedak et al. 1996; Arnbom et al. 1997; Carlini et al. 1997; Hindell & Slip 1997). Female foraging success is an important component of population status because it affects juvenile survival (McMahon et al. 000; Hall, McConnell & Barker 001). WEAN MASS Inter-island comparisons of wean mass provide a valuable tool for investigating the peculiar characteristics of the marine environment at each location (Burton et al. 1997). Wean mass remained unchanged during two periods of study at Macquarie Island suggesting that environmental conditions and/or the age distribution of foraging females have also remained constant. At Marion Island mean wean masses increased by approximately 5% over the same time which suggests an increase in the quality, quantity, or both, of prey or an increase in the ability of females to procure these prey. Vergani, Stanganelli & Bilenca (001) lend support to this view and demonstrated differences in elephant seal mean weaning mass that was believed to be a consequence of food availability subject to environmental perturbations caused by El Niño. Alternatively, the increase in wean mass at Marion Island may have occurred as a result of an increase in the mean age of females in the population, older females producing larger and heavier pups (Arnbom et al. 1993). The increase in wean mass at Marion Island could therefore be the outcome from an increase in the survival of older females rather than an increase in resources (see also Burton et al. 1997). PRIMIPARITY Marion Island females breed almost a year earlier than females at Macquarie Island. Because primiparity is related to body size and condition in seals (Laws 1956), we hypothesize that female seals at Marion Island are achieving adulthood (in terms of breeding) at a younger age through faster growth rates especially in capital breeders (animals that do not feed during the nursing period) such as elephant seals (Boyd 000). Age at first breeding at Marion decreased from 4 4 years (Bester & Wilkinson 1994) to 3 6 years and at Macquarie Island it decreased from 5 years (Hindell 1991) to 4 68 years. This reduction in primiparity may indicate that seals at both locations are currently showing improved growth either as a consequence of increased resource availability in the seal s Southern Ocean foraging grounds or a relative per capita increase in prey resources to the survivors. Furthermore, changes in reproductive parameters often occur after changes in population size that are caused by changes in food availability and climate change (Bowen, Capstick & Sergeant 1981; Lunn, Boyd & Croxall 1994), as has been mooted for the Marion Island population (Pistorius et al. 001). A reduction in age at first breeding may also be the consequence of individuals responding to selection pressure when densities are low (Stearns 1983). However, this seems unlikely because in longlived species such as elephant seals there is a tradeoff between current reproductive output and future reproductive success, unlike small mammals that respond by investing wholly in current offspring (Stearns 1983). Female elephant seals that breed before attaining a threshold size would suffer greater mortality and so compromise their fitness, which seems contrary to the life-history predictions for large long-lived mammals.

10 70 C. R. McMahon, H. R. Burton & M. N. Bester 003 British Ecology, 7, Greater resource acquisition can occur through either: (1) the resources being more available, () the females being more adept at acquiring these resources or (3) the resources being closer so that less energy is expended when the seals travel to and from foraging grounds. Evidence for clear differences in foraging behaviour of elephant seals of comparable ages and reproductive status in these populations is lacking, although adult females of the two populations have thus far showed no overlap (Hindell et al. 1994; Jonker & Bester 1998). There is, however, evidence of differences in productivity, and consequently resource availability, within the Southern Ocean (Smith, Stammerjohn & Baker 1996; Smith, Baker & Stammerjohn 1998; Wilson et al. 001). Faster growth and earlier attainment of breeding size at the small Marion Island population compared to the large Macquarie Island population suggests greater per capita prey availability. The increased reproductive rate would consequently be responsible for increased population growth (Huber, Beckham & Nisbet 1991) and, in the case of Marion Island, would have contributed to the possible stabilization of the population reported earlier (Pistorius et al. 001). However, the stabilization of the Marion Island population over the period (Pistorius et al. 001) has been questioned (Bradshaw et al. 00). Indeed, the lower net reproductive ratios presented here (for the period ) appear to support the conclusions of Bradshaw et al. (00). The rate of population change (decrease) at Marion Island ( 4 3%) is greater than that at Macquarie Island ( 1 47%, AAD data), but the differences in R 0 appear not to wholly justify the differences in population growth rates, as the rates of population change appear to be disproportionately high in relation to the relatively small differences in R 0. Similar observations have been demonstrated previously (Brommer 000). Consequently it appears that changes (increase in this instance) in fecundity have a limited effect on the rate of population change (Bester & Wilkinson 1994), and that the stabilization of the Marion Island population (Pistorius et al. 001) remains an open question. EMIGRATION AND PREDATION PRESSURE Assuming that per capita prey abundance is more plentiful for the Marion Island population, the greater survival of juveniles (both sexes) and early adults (females only) at Macquarie Island demands explanation. If the seals at Marion Island were growing faster and there were more resources available to them, their survival should be correspondingly higher unless there were more factors than resource limitation alone driving survival. Other factors such as predation and/or disease (Hindell et al. 1994) may be implicated. Very little permanent migration is possible, granted the genetic differences (Slade et al. 1998) of the populations at either Macquarie Island or Marion Island; and little emigration has been observed despite intensive beach searches at many sites (Nicholls 1970; Condy 1978; Bester 1988; Guinet et al. 199; Hindell & McMahon 000). Therefore migration from these islands is not likely to contribute to differences in survival. Killer whales (Orcinus orca, Linnaeus 1758) are known predators of southern elephant seals (Condy, van Aarde & Bester 1978; Guinet 199b; Keith et al. 001) and might play a significant localized role in the regulation of elephant seal populations (Condy et al. 1978; Hindell 1991; Guinet 199b). When populations are large (such as at Macquarie Island), the effects of killer whale predation on elephant seal population dynamics may appear minimal. However, at lower levels the effects are probably more significant and may be sufficient to drive a population decrease (Trites, Christensen & Pauly 1997). Similar numbers of killer whales occur at each island. At Macquarie Island where there are approximately 0 different whales (Copson 1994), and at Marion Island, where there are between 5 and 30 individuals (Pistorius et al. 00b). At Macquarie Island the ratio of killer whales to seals is : 1 while at Marion Island it is 0 10 : 1. Therefore, this considerable difference in predation pressure may account in part for the observed local differences in survival of juvenile elephant seals at Macquarie and Marion Islands. CAUSES FOR THE DIFFERENCES IN SURVIVAL AT MACQUARIE AND MARION ISLANDS: A HYPOTHESIS In summary, there are four important differences between the populations on Macquarie and Marion Islands. The population at Macquarie Island (compared to that at Marion Island): (1) is larger (5 8 times), () has an age at first breeding that is almost a year later, (3) showed no changes in the mean annual weanmasses for the last 7 years and (4) has less predation by killer whales. We interpret these differences to mean that: 1. Intraspecific competition at Marion Island was greater than at Macquarie Island and that even though there was more food, competition for it would have been greater. This is especially true in the first year of life when seals are naïve and for 3-year-old female seals that are making the physical and physiological change from juveniles to adults when we observe the lowest rates of survival. Intraspecific competition would act in such a way that juvenile seals, that forage in different parts of the water column to the adults (Hindell, Burton & Slip 1991a; Hindell, Slip & Burton 1991b; Slip 1997a,b; Hindell et al. 1999; Irvine et al. 000) are unable to compete with older seals as they, respectively: (1) enter a novel and foreign environment after weaning and () switch from juvenile foraging areas to adult foraging areas at the rapid onset of growth (Bryden 1968; Carrick, Csordas & Ingham 196; McLaren 1993). This necessarily brings experienced foragers and inexperienced foragers into direct competition for

11 71 Elephant seal life history and population change 003 British Ecology, 7, limited and patchily distributed prey (McConnell, Chambers & Fedak 199; McConnell, Fedak & Hunter 1993; McConnell & Fedak 1996). Indeed, Slip (1997b) has suggested that there is considerable overlap between the foraging areas of adult and juvenile seals from Heard Island. Although this (Slip 1997b) interpretation lends some support to our argument it needs to be treated cautiously because of the small sample sizes in the study. Younger seals do, however, develop and change their diving behaviour and by age juvenile northern elephant seals (Mirounga angustirostris, Gill 1866) have developed diving abilities similar to those of adult seals (LeBoeuf 1994; LeBoeuf et al. 1996). It thus appears conceivable that relatively inexperienced seals older than two (in the third year of life) are competing with older, more experienced seals for resources.. That there has been a general increase in the availability of resources in the Southern Ocean that allows females to grow faster and attain breeding size earlier, but that this increase has not been uniform around the Southern Ocean, there being a relatively greater availability of resources in the regions where the Marion Island seals feed. 3. That the increase in wean mass at Marion Island and the stability of the wean masses at Macquarie Island prior to 199 and after 1993 is proof of (a) an increase in resources or (b) an increase in the mean age of females at Marion Island. The latter is unlikely because, for this to occur, juvenile survival rates would have to decrease relative to adult survival or adult survival would have to increase relative to juvenile survival over these two periods. Neither appears to have occurred at Marion Island (Pistorius et al. 1999b; Pistorius et al. 001; this study). 4. That predation pressure is greater at Marion Island such that it may play a significant role in the mortality of seals at Marion Island to the extent that the lower overall rates of survival there might be attributed to killer whale predation. Conclusions Elephant seal population decreases in the southern Indian and Pacific Oceans appear to have been driven by resource limitations in the Southern Ocean. Sea-ice extent is one recognized determinant of primary production in the Antarctic region of the Southern Ocean (Loeb et al. 1997; Nicol et al. 000). Sea-ice decreased dramatically albeit not uniformly from the 1950s to the 1970s (de la Mare 1997) when southern elephant seal populations were decreasing most rapidly. This decrease in sea-ice has recently stabilized (de la Mare 1997), which suggests a corresponding stabilization in primary production as well. Such a stabilization of production at the lowest level of the food chain could be translated into stabilization of the entire food chain over time. The decrease in primiparity at Macquarie Island and at Marion Island suggests that present population decreases in both populations may ameliorate, particularly at Marion Island where seals are breeding almost a year earlier than at Macquarie Island. As female seals are pupping earlier at Marion Island and Macquarie Island; and because reproductive maturity can only be achieved after seals have reached a minimum size (Laws 1956), it follows that seals are likely to be growing faster if they reach breeding size earlier. We believe that more resources may be available to seals now than in the 1950s when elephant seal populations began decreasing. Perhaps they can now achieve this minimum size earlier in life. Recent observations (de la Mare 1997) of stabilizations in sea-ice around Antarctica appear to support our view of changes in food supply. Regional differences in survival are partly the result of local differences in predator numbers for populations with greater predation pressure suffer greater mortality. It is likely that small populations are more prone to the negative effects of predators than larger populations (Guinet 199a; Trites et al. 1997). Regional differences in productivity within the Southern Ocean may also dictate local differences in survival because of differences in prey availability as shown for Adélie penguins (Pygosocelis adeliae, Hombron & Jacquinot 1841) (Wilson et al. 001). However, because elephant seals range and forage over wide areas of the Southern Ocean (Hindell et al. 1991a; McConnell et al. 199; McConnell & Fedak 1996; Jonker & Bester 1998; Hindell & McMahon 000) and because physical and environmental anomalies are transported around the Southern Ocean (White, Chen & Peterson 1998; White & Peterson 1996) it is difficult to relate seal life-history variations to any localized physical or environmental anomaly. Therefore resource variability would affect all seal populations such that it would be difficult to attribute small population differences in survival to global environmental variability. However, environmental variability can be expected to drive large-scale population changes (Barbraud & Weimerskirch 001). Elephant seal population decreases in the southern Indian Ocean and the southern Pacific Ocean were therefore probably driven by large-scale environmental change and ocean productivity in a similar manner to that shown for emperor penguins (Aptenodytes forsteri, Gray 1844) (Barbraud & Weimerskirch 001). Inter-island (population) differences are almost certainly the result of a complex interplay of predator pressure, intraspecific competition for resources and differences in resource availability in general. Intraspecific competition for resources is likely to be exacerbated in years when resources are limited and parameters such as survival can be expected to vary greatly at such times. If our explanation, for earlier breeding at Marion Island within the last decade, is that increases in productivity in the Southern Ocean have become more available (to the population of Southern Elephant Seals breeding there) is correct, then this is a clear

12 7 C. R. McMahon, H. R. Burton & M. N. Bester 003 British Ecology, 7, prediction to test. Data from satellite imagery of ocean colour and changes in the locations and strengths of ocean currents could be investigated with this hypothesis in mind. Acknowledgements We are extremely grateful to Mark Hindell for his thoughts and discussion of the data presented here and his valuable and considered review of earlier drafts of the manuscript. Pierre Pistorius and Steve Kirkman collated the Marion Island data and Dave Watts was responsible for the management of the Macquarie Island data. Anne York is acknowledged for her generous help in comparing the ages at first breeding. An anonymous reviewer is also thanked for his/ her positive criticism. Our numerous field assistants at Macquarie Island and Marion Island between 1993 and 001 are thanked for their tireless efforts to mark and resight seals. This study was logistically and financially supported by the South African Department of Environmental Affairs and Tourism on advice of the South African Committee for Antarctic Research (SACAR) and the Australian Antarctic Division through the Australian Nation Antarctic Research Expeditions (ANARE). This study was carried out at Macquarie Island under ethics approval from the Australian Antarctic Animal Ethics Committee (ASAC 65) and the Tasmanian Parks and Wildlife Service. The Ethics Committee of the Faculty of Natural and Agricultural Sciences of the University of Pretoria endorsed the research at Marion Island (reference number EC ) under permit from the Director-General: Environmental Affairs and Tourism. References Arnbom, T., Fedak, M.A. & Boyd, I.L. (1997) Factors affecting maternal expenditure in southern elephant seals during lactation. Ecology, 78, Arnbom, T.A., Fedak, M.A., Boyd, I.L. & McConnell, B.J. (1993) Variation in weaning mass of pups in relation to maternal mass, post-weaning fast duration, and weaned pup behaviour in southern elephant seals (Mirounga leonina) at South Georgia. Canadian Journal of Zoology, 71, Barbraud, C. & Weimerskirch, H. (001) Emperor penguins and climate change. Nature, 411, Barker, R. (199) Effect of heterogeneous survival on birdbanding model confidence interval coverage rates. Journal of Wildlife Management, 56, Bester, M.N. (1988) Marking and monitoring studies of the Kerguélen stock of southern elephant seals Mirounga leonina and their bearing on biological research in the Vestfold Hills. Hydrobiologia, 165, Bester, M.N., Möller, H., Wium, J. & Enslin, B. (001) An update on the status of southern elephant seals on Gough Island. South African Journal of Wildlife Research, 31, Bester, M.N. & Wilkinson, I.S. (1994) Population ecology of southern elephant seals at Marion Island. Elephant Seals: Population Ecology, Behavior, and Physiology (eds B.J. LeBoeuf & R.M. Laws), pp University of California Press, Berkeley. Bowen, W.D., Capstick, C.K. & Sergeant, D.E. (1981) Temporal changes in the reproductive potential of female harp seals (Pagophilus groenlandicus). Canadian Journal of Fisheries and Aquatic Science, 38, Boyd, I.L. (000) State-dependent fertility in pinnipeds: contrasting capital and income breeders. Functional Ecology, 14, Boyd, I.L., Walker, T.R. & Poncet, J. (1996) Status of southern elephant seals at South Georgia. Antarctic Science, 8, Bradshaw, C.J.A., McMahon, C.R., Hindell, M.A., Pistorius, P.A. & Bester, M.N. (00) Do southern elephant seals show density dependence in fecundity? Polar Biology, 5, Brommer, J. (000) The evolution of fitness in life-history theory. Biological Reviews, 75, Bryden, M.M. (1968) Development and growth of the southern elephant seal (Mirounga leonina) (LINN). Papers and Proceedings of the Royal Society of Tasmania, 10, 5 3. Burnham, K.P., Anderson, D.R., White, G.C., Brownie, C. & Pollock, K.H. (1987) Design and analysis methods for fish survival experiments based on release recapture. American Fisheries Society Monograph, 5, Burton, H. (1986) A substantial decline in numbers of the southern elephant seal at Heard Island. Tasmanian Naturalist, 86, 4 8. Burton, H.R., Arnbom, T.L.B.I., Bester, M.N., Vergani, D. & Wilkinson, I. (1997) Significant differences in the weaning mass of southern elephant seals from five sub-antarctic islands in relation to population declines. Antarctic Communities: Species Structure and Survival (eds B. Battaglia, J. Valencia & D.W.H. Watton), pp Cambridge University Press, Cambridge. Campagna, C., Lewis, M. & Baldi, R. (1993) Breeding biology of southern elephant seals in Patagonia. Marine Mammal Science, 9, Carlini, A.R., Daneri, G.A., Marquez, M.E.I., Soave, G.E. & Poljak, S. (1997) Mass transfer from mothers to pups and mass recovery by mothers during the post-breeding foraging period in southern elephant seals (Mirounga leonina) at King George Island. Polar Biology, 18, Carothers, A.D. (1979) Quantifying unequal catchability and its effect on survival estimates in an actual population. Ecology, 48, Carrick, R., Csordas, S.E. & Ingham, S.E. (196) Studies on the southern elephant seal Mirounga leonina (L.). IV. Breeding and development. CSIRO Wildlife Research,, Carrick, R. & Ingham, S.E. (1960) Ecological studies of the southern elephant seal Mirounga leonina (L.), at Macquarie Island and at Heard Island. Mammalia, 4, Carrick, R. & Ingham, S.E. (196) Studies on the southern elephant seal Mirounga leonina (L.). I. Introduction to the series. CSIRO Wildlife Research, 7, Caughley, G. (1977) Analysis of Vertebrate Populations. John Wiley and Sons, London. Chittleborough, R.G. & Ealey, E.H.M. (1951) Seal marking at Heard Island, ANARE Interim Report 1 (ed. P.G. Law), pp Antarctic Division, Department of External Affairs, Melbourne. Condy, P.R. (1978) The distribution and abundance of southern elephant seals Mirounga leonina (Linn) on the Prince Edward Islands. South African Journal of Antarctic Research, 8, Condy, P.R., van Aarde, R.J. & Bester, M.N. (1978) The seasonal occurrence and behaviour of killer whales (Orcinus orca) at Marion Island. Journal of Zoology, London, 184, Copson, G.R. (1994) Cetacean sightings and strandings at subantarctic Macquarie Island, Report no. 91. Australian Antarctic Division, Hobart.